ebook img

Assessing Wildland Fire Risk Transmission to Communities in Northern Spain PDF

27 Pages·2017·5.56 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Assessing Wildland Fire Risk Transmission to Communities in Northern Spain

Article Assessing Wildland Fire Risk Transmission to Communities in Northern Spain FermínJ.Alcasena1,*,MicheleSalis2,AlanA.Ager3,RafaelCastell4and CristinaVega-García1,5 1 AgricultureandForestEngineeringDepartment(EAGROF),UniversityofLleida,AlcaldeRoviraRoure191, Lleida25198,Spain;[email protected] 2 Euro-MediterraneanCenteronClimateChange(CMCC),IAFESDivision,ViaEnricoDeNicola9, Sassari07100,Italy;[email protected] 3 USDAForestService,RockyMountainResearchStation,72510CoyoteRoad,Pendleton,OR97801,USA; [email protected] 4 BomberosdeNavarra,AgenciaNavarradeEmergencias(ANE),CalleAoiz35bis3◦,Pamplona31004, Spain;[email protected] 5 ForestSciencesCentreofCatalonia,CarreteradeSantLlorençdeMorunyskm2,Solsona25280,Spain * Correspondence:[email protected];Tel.:+3-497-370-2673 AcademicEditors:DaveVerbylaandTimothyA.Martin Received:20November2016;Accepted:18January2017;Published:24January2017 Abstract: We assessed potential economic losses and transmission to residential houses from wildland fires in a rural area of central Navarra (Spain). Expected losses were quantified at the individualstructurelevel(n=306)in14ruralcommunitiesbycombiningfiremodelpredictions ofburnprobabilityandfireintensitywithsusceptibilityfunctionsderivedfromexpertjudgement. Fireexposurewasestimatedbysimulating50,000fireeventsthatreplicatedextreme(97thpercentile) historical fire weather conditions. Spatial ignition probabilities were used in the simulations to accountfornon-randomignitions,andwereestimatedfromafireoccurrencemodelgeneratedwith anartificialneuralnetwork. Theresultsshowedthatignitionprobabilityexplainedmostofspatial variationinrisk,witheconomicvalueofstructureshavingonlyaminoreffect. Averageexpectedloss toresidentialhousesfromasinglewildfireeventinthestudyareawas7955€,andrangedfromalow of740tothehighof28,725€. Majorfireflow-pathswereanalyzedtounderstandfiretransmission fromsurroundingmunicipalitiesandshowedthatincomingfiresfromthenorthexhibitedstrong pathwaysintothecoreofthestudyarea,andfiresspreadingfromthesouthhadthehighestlikelihood ofreachingtargetresidentialstructuresfromthelongestdistances(>5km). Communityfiresheds revealed the scale of risk to communities and extended well beyond administrative boundaries. Theresultsprovidedaquantitativeriskassessmentthatcanbeusedbyinsurancecompaniesandlocal landscapemanagerstoprioritizeandallocateinvestmentstotreatwildlandfuelsandidentifyclusters ofhighexpectedlosswithincommunities. Themethodologicalframeworkcanbeextendedtoother fire-pronesouthernEuropeanUnioncountrieswherecommunitiesarethreatenedbylargewildlandfires. Keywords: wildland urban interface; wildfire simulation modeling; wildfire risk transmission; communityfireshed 1. Introduction Mostwildfiresthatcausehumanfatalitiesandlossestopropertyoccurintherapidlyexpanding interfaceareasbetweenwildlandsandhumandevelopment[1,2]. Thisareawhereresidentialand other infrastructures intermingle with flammable vegetation is widely known as wildland–urban interface (WUI) or rural–urban interface (RUI) [3,4]. While the former definition is mainly used Forests2017,8,30;doi:10.3390/f8020030 www.mdpi.com/journal/forests Forests2017,8,30 2of27 for predominantly wildland vegetation areas surrounding developed areas, the latter is most commonlyusedinMediterraneanlandscapeswherefuelshavebeeninfluencedbyhumanactivities formillennia[5,6]. Inareaslackingsharptransitionsbetweendevelopmentandwildlands, where structures are surrounded by hazardous fuels, the term intermix has been used to describe the juxtapositionoffuelsanddwellings[7]. Inallcases,anumberoffactorshavecontributedtowildfire lossesindevelopedareas(hereafterWUI),includingurbanexpansion,increasedfuelloadingsfrom expansionofshrubandforestvegetationintoabandonedagriculturallands,andsuburbansprawl overmetropolitanagriculturalbelts[8–10]. Likewise,firesuppressionpolicieshavecontributedtoa buildupoffuelsinandarounddevelopedareas,resultinginhigherhazardwithindevelopedareas andstructureignition[11]. WildlandfiresintheWUIareagrowingconcernatglobalscalesdueto escalatinglossestolifeandproperty[12,13],andhavebecomeapriorityforwildfiremanagement policiesinmanyfire-proneareas[14]. Previous efforts on WUI wildfire risk characterization in Mediterranean landscapes have emphasizedtheimportanceofflammablevegetationsurroundingcommunities[15],sincefuelloadings aredirectlyrelatedtofireintensityandstructureloss[16]. Aggregationofdwellings(isolated,grouped and urban center) combined with vegetation types or land covers have been proposed as a WUI classificationsystemtoinformriskandvulnerabilityassessments[17,18]. Otherstudieshavefocused onignitionlikelihoodtomeasurewildfirerisk[19–21]. ThevastmajorityoffiresintheMediterranean basinarecausedbyhumans[22–24],andmostfire-occurrencemodelingstudiesincludeexplanatory variablestodescribehumanactivities,suchaspopulationdensity,accessibility(e.g.,distancetoroads, distancetorailways,distancetoforesttracks)andhumanactivities[25,26]. However,neitherofthese previousapproachesaccountforthelikelihoodoflossfromlargefires(e.g.,5000–50,000ha)thatignite atsomedistantlocationandspreadtourbandevelopment. Thus,lowfireignitionprobabilityclose toaWUIareadoesnotnecessarilytranslatetolowburnprobability,andviceversa. Moreover,fire intensitycansubstantiallyvarydependingonfireweatherandfirefrontspreadingdirection[27,28]. Tobetteraccountforthespatialscaleofwildfirerisktohumancommunities,agrowingnumber ofresearchershaveemployedwildfiresimulationmethods[29,30]. Bothburnprobabilityandfire intensityinthehomeignitionzone(HIZ,theimmediate30–60m-bufferareaarounddwellings)[11] canbeestimatedbysimulatingalargenumberoffires(e.g.,104–105)toassesswildfireexposurefrom large fires [31,32]. These estimates can be then used in risk assessments to quantify the potential socioeconomic impacts, including expected net value change on residential structures [27,28,33]. WhileitisgenerallyagreedthathigherwildfireexposureresultsinlargerlossesintheWUI,variability instructuresusceptibilityandeconomicvaluationcansubstantiallyaffectriskestimatesatthescale ofindividualdwellings. Forinstance,highoverallexposurelevelscanbemitigatedbyconstruction materialsandstructuredesign[34]. Thesedifferencesinconstructioncanbeincorporatedintorisk assessmentsusingdifferentsusceptibilityrelationships[35,36]. Simulationstudiescanalsobeused to understand the scale of risk to communities to help identify responsible landowners [37,38]. For instance, using wildfire transmission analysis, fire effects on valued resources can be traced backtotheignitionlocation[39],andlandscapeplanningtoreducehazardousfuelscanthentarget theseareasforfueltreatments[40]. Inthispaper,weassesspotentialwildfireeconomiclossesandtransmissiontoresidentialhouses intheruralcommunitiesofJuslapeñaValley,northernSpain. Weusedsimulationmodelingtomap thesourceofwildfireexposuretocommunitiesandestimatedtheexpectedfinanciallossatthescale ofindividualstructures. Thesimulationmodelingincorporatedafinescaleignitionprobabilitygrid developed from historical fire locations. Simulation outputs were used to estimate a number of exposuremetrics,includingburningprobabilityandfireintensity. Weestimatedexpectedlossinthe communityusingwildfireexposuremetricscombinedwithastructuresusceptibilityfunction.Thelater wasgeneratedbyapaneloflocalexpertsusinganinteractivestructuredcommunicationtechnique. We also conducted a transmission analysis to delineate community firesheds and understand the sourceofwildfireexposuretocommunities. Themethodsprovideanumberofnewwaystoexamine Forests2017,8,30 3of27 wildfireexposuretocommunitiesthatcaninformwildfireprotectionandimprovefireresiliencyin rural–urbaninterfaceareasintheMediterraneanregion. 2. MaterialandMethods Forests 2017, 8, 30   3 of 29  2.1. StudyArea interactive structured communication technique. We also conducted a transmission analysis to  Thestudyadreelinaeaitse cloomcmatuendityi fniretshheedsJ aunsdl aunpdeerñstaanVd athlel esoyu,rccee onf twraildlfiNre aexvpaorsurrae t(oS cpomaimnu)n,it1ie8s. kmnorthofthe cityofPamplonaTh(eF migetuhordes 1pArov).idTe ha enuJmubselra opf neeñwa wVayasl lteo yexaismiane3 w1i.l6d3firke emxp2osmurue tno iccoimpmaulintiytiews thitaht  548inhabitants can inform wildfire protection and improve fire resiliency in rural–urban interface areas in the  dispersedamongMe1d4itesrmranaelalnr ruegriaonl. villagesorcouncils(minimumadministrativedivision). Theclimate istransitionalMediterraneanwithannualrainfallaround1000mm, awatershortageperiodfrom 2. Material and Methods  JulytoSeptembercorrespondingtothewildfireseason,andaveragemaximumtemperaturesover 30◦Cinthewar2m.1. eSstutdmy Aorena th(meteo.navarra.es). Thelandscapeisamosaicofdrylandcerealcrops The study area is located in the Juslapeña Valley, central Navarra (Spain), 18 km north of the  coveringthevalleybottom,mesoxerophyticpastureswithshrubbyedgingsonmarginalagricultural city of Pamplona (Figure 1A). The Juslapeña Valley is a 31.63 km2 municipality with 548  lands(GenistascoinrhpaibuitsanLts. ,dJiuspnerispeedr uamsocnogm 1m4 usmnaisll Lru.,raBl uvixlluagsess eomr pcoeurnvciirlse n(msiLni.m,aumn dadPmriuninsturastivsep inosaL.),downy oak(Quercuspubdeivsicsieonn)s. TMhei clllim.)atfeo isr terasntssitioonnals Moeudtitherrfaanceainn wgitshl aonpnueasl r(arinefpalll aarcoeudndb 10y00Q mume, rac wuasteirl exL.inshallow shortage period from July to September corresponding to the wildfire season, and average  soilfoothills),bemeacxhim(uFma tgemuspesraytluvreast oicvaer L30. )°Cf oinr tehset wsaormnesht imgohnthe l(mevetaeot.inoanvarnrao.erst).h Thfea lcainndgscaspleo isp ae s,andscattered standsofblackpmionseaic( Poif nduryslannidg rceareAal rcnro.)ps[ 4c1ov].erLinag nthde mvaallenya bgoettmome, nmteissoxlearorpgheyltyic cpoasntudreist iownithe dbyownership. shrubby edgings on marginal agricultural lands (Genista scorpius L., Juniperus communis L., Buxus  Mostforestsandnaturalherbaceouspasturesarecouncilcommonlandsandagriculturalfieldsare sempervirens L., and Prunus spinosa L.), downy oak (Quercus pubescens Mill.) forests on south facing  ownedbylocalsilnopheas b(rietpalancetds .byC Qoumercmus uilenx iLt. yin hshoaulloswin sgoil ifsooltohicllas)t, ebdeecaht (Fmaguids s-yslvloatpicae Ls.,) fuorseustas lolny  surroundedby high elevation north facing slopes, and scattered stands of black pine (Pinus nigra Arn.) [41]. Land  agricultural lands and orchards at the front southern side, and forested lands arrive closer at the management is largely conditioned by ownership. Most forests and natural herbaceous pastures  back(Figure1B)a.rWe ceoufnocicl ucosmemdoon ularndasn aandly asgirsicuolnturrael sfiiedldes nartei aolwhneodu bsye lsoc(anl in=ha3b0it6anstst. rCuocmtmuurneisty) ,andwedidnot considerothersthrouucstinugr eiss loocartecdo ant smtridu‐sclotipoesn, sussuualclyh suarsroaugndriecdu blyt uagrraicluwltuararle lahnodus saneds .orIcnhatrhdse ast ttuhed yarea,thereare front southern side, and forested lands arrive closer at the back (Figure 1B). We focused our  noindustrialsiteasnaolyrsiss pono rrets-idreenctriael ahtoiuosnesa (ln f=a 3c0i6l isttriuecst.urTesh), eanlda rwgee dsitd onobt sceonrvsideedr owthielrd stfirurcetusreasr oer characterizedas fast-spreadingoncoen-sdtrauyctiosnusm sumch ears aegvriecunlttusrawl witahrehleousssest.h Ina nthe1 s0tu0d0y haraeab, tuhrernee adre (neo. ign.d,uJsutrsialla spiteesñ oar Firein2009and sport‐recreational facilities. The largest observed wildfires are characterized as fast‐spreading  SanCristobalFirein2001). Mostfiresarecausedbyhumans,whilelightningrepresentsonly5%of one‐day summer events with less than 1000 ha burned (e.g., Juslapeña Fire in 2009 and San  ignitions(1985toCr2is0to1b3al fiFirree irne 2c0o01r)d. Ms;osmt fairpesa amre aca.gusoedb .beys h)u.mans, while lightning represents only 5% of  ignitions (1985 to 2013 fire records; mapama.gob.es).    Figure 1. Lo cationoftheJuslapeñaValley(3163ha)incentralNavarra(Spain)(A).Thenumbers refer to the regional cadaster council polygon (1 to 16) and municipality code (B). The 36,000-ha wildfiremodelingdomainframedbythelandscapefile(LCP)encompassingthestudyareahada widerextensiontothesouthtoaccountforincomingfiresfromthefire-proneareasofcentralNavarra. Landcoversintheculturallandscapespresentsharpedgesinvegetation(Burbancenterofthecouncil No.8).Detailedcartographyandcadasterpolygons(scale1/5000)wereusedtogeneratesurfacefuel maps(sigpac.navarra.es)andlocateresidentialhouses(catastro.navarra.es)(C).TheHIZisthe60-m bufferaroundstructures[11],andwascalculatedforeachresidentialhousetoconductthisstudy.Inthe figure(C)weshowtheexternalHIZcontouroftheresidentialhousesinurbancenterofthecouncilNo.8. Forests 2017, 8, 30   4 of 29  Figure 1. Location of the Juslapeña Valley (3163 ha) in central Navarra (Spain) (A). The numbers  refer to the regional cadaster council polygon (1 to 16) and municipality code (B). The 36,000‐ha  wildfire modeling domain framed by the landscape file (LCP) encompassing the study area had  a wider extension to the south to account for incoming fires from the fire‐prone areas of central  Navarra. Land covers in the cultural landscapes present sharp edges in vegetation (B urban center  of the council No. 8). Detailed cartography and cadaster polygons (scale 1/5000) were used to  generate surface fuel maps (sigpac.navarra.es) and locate residential houses (catastro.navarra.es)  (C). The HIZ is the 60‐m buffer around structures [11], and was calculated for each residential  Forests2017,8,30 4of27 house to conduct this study. In the figure (C) we show the external HIZ contour of the residential  houses in urban center of the council No. 8.  2.2. WildfireSimulation 2.2. Wildfire Simulation  We gathWeree gdatmheureldti mplueltdipalet adsaettasseatsn adndg geeoossppaattiiaall ininpputust fsorf othrist hmiosdmeliondg ealpipnrgoaacph p(Friogaucrhe 2()F. igure 2). WesimulWatee dsimwuilldatfierde wspilrdefairde asnpdreabde haanvdi obreh(fiarveiosri z(efi,reb usirznee, dbuarrneeadp oarleyag opnosly,gfloanms, efllaemnge thlenpgrtohb abilities, probabilities, and conditional burn probability) within a 36,000‐ha fire modeling domain. Overall,  andconditionalburnprobability)withina36,000-hafiremodelingdomain. Overall,weconducted we conducted separated simulations for the most frequent extreme weather conditions of the  separatedsimulationsforthemostfrequentextremeweatherconditionsofthewildfireseason,thus wildfire season, thus obtaining different sets of modeling outputs. All the output raster grids  obtainingdifferentsetsofmodelingoutputs. Alltheoutputrastergridswereobtainedatmodeling were obtained at modeling resolution. Details are presented below in the following sections  resolutio(nT.abDlee t1a).i lsarepresentedbelowinthefollowingsections(Table1).   Figure2.FWiguilrde fi2r.e Wsiimldfuirlea tsioimnualantdiona naanldy sainsaplyrsoisc epsrsocseusms smumarmyaflryo wflocwhachrta.rtW. Wildilfidrfieres ismimuullaattiioonn requires requires fire weather, landscape and fire ignition input data. Fire initiation, transmission,  fireweather,landscapeandfireignitioninputdata. Fireinitiation,transmission,exposureandrisk exposure and risk analysis use different fire modeling outputs. Exposure and risk analyses were  analysisusedifferentfiremodelingoutputs.Exposureandriskanalyseswereconductedatindividual conducted at individual structure HIZ level. Results were presented in maps or graphics. See  structureHIZlevel.Resultswerepresentedinmapsorgraphics.SeeTable1fortheabbreviations. Table 1 for the abbreviations.  Table1. Summarytablewiththeabbreviationsusedinthisstudyforthemaingeospatialinputs, modelingoutcomesandanalysisresults.Thetermsaredescribedandcontextualizedfortheuseinthis study.Weprovidefurtherdetailsinthefollowingsections. Name(Abbreviation) DescriptionandUse Areasurroundingstructureswithina30–60m-buffer[11].HIZwasused Homeignitionzone(HIZ) toassesswildfireexposureandriskontheindividualresidentialhouses locatedinthestudyarea. Fireoccurrenceprobabilitygrid(0–1)generatedbyartificialneural Ignitionprobability(IP) networkanalysis[26]ofhistoricalignitionlocations.Itwasusedto calculateFPIandgeneratethesimulatedfireignitionlocations. Firesize(ha)resultingfromeachindividualsimulatedwildfire.Firesize Firesize(FS) isoutputfromsimulationsalongwiththeignitionlocation.Itwas combinedwithIPtogenerateFPI. Forests2017,8,30 5of27 Table1.Cont. Name(Abbreviation) DescriptionandUse IsthegridgeneratedwithFSandIP,anditwasusedtoidentifylargefire initiationareas[31].TheFPIprovidesspatiallyexplicitvaluable Firepotentialindex(FPI) informationtotargetanthropicfireignitionpreventionpriorityareason fire-pronelandscapes. Probabilityofafireofaspecificflamelengthgiventhatapixelburns underthesimulatedconditions.FLPisoutputin0.5-mclassesandsums Flamelengthprobability(FLP) to1foragivenpixel.Adistributionofflamelengthsisgeneratedforeach pixelsincefirescanarriveasheading,flankingorbackingfires. Probability-weightedflamelength(m)calculatedfromtheFLPoutput. Conditionalflamelength(CFL) CFLwassummarizedfortheHIZtoestimatewildfirehazardand exposuretoresidentialhouses[35]. Numberoftimesapixelburnsasaproportionofthetotalnumberof simulatedfires(0–1).BPaveragevaluesforeachHIZwereusedto Burnprobability(BP) estimatewildfirelikelihoodandassesswildfireexposuretoresidential houses[35]. Thesusceptibilityofstructuresasafunctionofflamelengthrepresented Responsefunction(RF) bypercentvalueloss(%)[42].Itwasobtainedfromexpertjudgment[35]. Expectationofgainorlossinvaluesexpressedonapercentagebasis (%)[28].Derivedfromcombiningburnprobability,intensity,and Expectednetvaluechange(eNVC) susceptibilityfunctionstoestimateexpectedchangeonapercentagebasis forstructures[27].Onlyexpectedlosseswereconsideredinthestudy. Expectedlossexpressedspecificallyineconomicvalues(€)givenafire ignitionandspreadatassumedextremefireweatherconditions. Expectedeconomicloss(eEL) Quantifiedastheproductofthecadastervalueofthestructuresandthe averageeNVCwithintheHIZ. 2.2.1. LandscapeFileandFireWeatherInputData We compiled the complete set of input data as required by the FlamMap fire simulator [43], includinglandscapefile(LCP)andwildfireseasonextremefireweatherdata. TheLCPisagridded framecontainingthecharacteristicsoftheterrain,surfacefuelsandcanopyfuelmetrics. Theterrain (aspect,slopeandelevation)wasderivedfrom5-mresolutiondigitalterrainmodelrasterdata(ign.es). Standardfuelmodels[44,45]wereassignedto1/5000scalelanduselandcoverconsideringspecies composition,shrubcoverandforestgrowthstage(idena.navarra.esandsigpac.navarra.es)(Figure3). Canopymetrics(canopyheight,canopybaseheight,canopybulkdensityandcanopycover),were derivedfromlowdensityLiDARdata(0.56returnsm2;ign.es)usingFUSION[31,46]. Thesurface fuelandcanopymetriccharacterizationandrequiredrastergridgenerationweredetailedinprevious studies [47,48]. The LCP was assembled at 20-m resolution [49] and comprised a 36,000-ha fire modeling domain (Figure 1A). Extreme fire weather conditions were derived using Fire Family Plus[50]fromthehourlyrecordsofthePamplonaautomaticweatherstation(1999to2015records; meteo.navarra.es), as the 97th percentile ERC-G fuel moisture content [51] and wildfire season dominant winds (Table 2). We generated five wind scenarios considering the most frequent wind directions(frequency>5%inweatherrecords)duringwildfireseasonandtherespective97thpercentile windspeeds. Forests 2017, 8, 30   6 of 29  We compiled the complete set of input data as required by the FlamMap fire simulator [43],  including landscape file (LCP) and wildfire season extreme fire weather data. The LCP is a  gridded frame containing the characteristics of the terrain, surface fuels and canopy fuel metrics.  The terrain (aspect, slope and elevation) was derived from 5‐m resolution digital terrain model  raster data (ign.es). Standard fuel models [44,45] were assigned to 1/5000 scale land use land cover  considering species composition, shrub cover and forest growth stage (idena.navarra.es and  sigpac.navarra.es) (Figure 3). Canopy metrics (canopy height, canopy base height, canopy bulk  density and canopy cover), were derived from low density LiDAR data (0.56 returns m2; ign.es)  using FUSION [31,46]. The surface fuel and canopy metric characterization and required raster  grid generation were detailed in previous studies [47,48]. The LCP was assembled at 20‐m  resolution [49] and comprised a 36,000‐ha fire modeling domain (Figure 1A). Extreme fire  weather conditions were derived using Fire Family Plus [50] from the hourly records of the  Pamplona automatic weather station (1999 to 2015 records; meteo.navarra.es), as the 97th  percentile ERC‐G fuel moisture content [51] and wildfire season dominant winds (Table 2). We  Forgesetns2e0r1a7t,e8d,3 f0ive wind scenarios considering the most frequent wind directions (frequency >5%6 oinf2 7 weather records) during wildfire season and the respective 97th percentile wind speeds.    Figure 3. Land cover map (idena.navarra.es) and assigned fuel models [44,45] for the wildfire  Figure3.Landcovermap(idena.navarra.es)andassignedfuelmodels[44,45]forthewildfiremodeling. modeling. The large urban development areas in the southeast correspond to the capital city of  The large urban development areas in the southeast correspond to the capital city of Pamplona. Pamplona. Cereal crops occupy all the flat cultivated areas to the south and mountains in the  Cerealcropsoccupyalltheflatcultivatedareastothesouthandmountainsinthenorthernpartare northern part are covered by mosaics of different forest types. See fuel model parameter details  coveredbymosaicsofdifferentforesttypes.Seefuelmodelparameterdetailsinthereferences[44,45]. in the references [44,45].  Table2.Fireweatherinputdata,correspondingtothehistorical97thpercentileconditions,usedfor Table 2. Fire weather input data, corresponding to the historical 97th percentile conditions, used  wildfiresimulations. Weconsideredthemostfrequentwinddirections(frequency>5%)duringthe for wildfire simulations. We considered the most frequent wind directions (frequency >5%)  last17wildfireseasons. Historicalweatherdataweregatheredfromthemeteorologicalstationof during  the  last  17  wildfire  seasons.  Historical  weather  data  were  gathered  from  the  Pamplona(meteo.navarra.es).Weusedstandardfuelmodelsforfiremodeling,seereferences[44,45] meteorological station of Pamplona (meteo.navarra.es). We used standard fuel models for fire  forfurtherdetails. modeling, see references [44,45] for further details.  WindScenario FuelMoistureContent(%) Wind Scenario  Fuel Moisture Content (%)  Direction (°)  Probability FFuueellM Mooddeell[ 4[444,4,455]]  Direction(◦) (kSmp·ehe−d1) Probability FuCelatLeogaodriyng GGRS12,,GGRR54,, TU3,PCL, GR1,SH3, SH3,GR3 TL2,SH8 SH6,SH5 67.5 32 0.43 1-h 4 6 8 337.5 35 0.28 10-h 5 7 9 45.0 19 0.17 100-h 8 9 12 180.0 31 0.06 Liveherbaceous 20 45 70 22.5 23 0.06 Livewoody 60 85 100 GS1=lowload,dryclimategrass-shrub;GR2=lowload,dryclimategrass;GR4=moderateload,dryclimategrass; GR5=lowload,humidclimategrass;SH6=lowload,humidclimateshrub;SH5=highload,dryclimateshrub; TU3=moderateload,humidclimatetimber-grassshrub;PCL=closedandlowlitterpinestands;SH3=moderate load,humidclimateshrub;GR3=lowload,verycoarse,humidclimategrass;GR1=short,sparsedryclimategrass; SH3=Moderateload,humidclimateshrub;TL2=lowloadbroadleaflitter;SH8=highload,humidclimateshrub.” 2.2.2. FireOccurrenceModeling Weusedartificialneuralnetworks(ANNs)toconstructafireoccurrencemodel,andultimatelyto generatea20-mresolutionignitionprobabilitygridencompassingthemodelingdomain. A10,000-fire ignitionpointinputfileforwildfiresimulationwasthencreatedfromtheignitionprobability(IP) grid masked to burnable fuels. ANN models are robust pattern detectors which can approximate mathematicalrelationshipswithnon-normaldistributionsandspatiallycorrelatedvariableswhere Forests2017,8,30 7of27 otherstatisticalmodelscouldcausemulticollinearity[52,53],andhavebeensuccessfullyappliedto fireoccurrencepredictioninpreviouswork[26,54]. Thehistoricalfireignitionswithinthefiremodelingdomain(200ignitionsinall,1985to2013 firerecords; mapama.gob.es)andthesamenumberofrandomno-fireobservationswerematched to topography (elevation, aspect, slope), land cover class, population density, and accessibility (distancetoroads,tracks,railways,urbanareasandpowerlines)20-mresolutionrastergrids(ign.es; idena.navarra.es). Ten percent of the fire and no-fire observation variable dataset (40 cases) was set apart for validation purposes before model building. We selected feed-forward, multilayered, non-linear, fully connected, cascade-correlation networks [55], built using Neural Works Predict® v.3.30software(NeuralWorksPredict®3.30,SerialNumberNPSC30-70755,Carnegie,PA,USA)[56] withanadaptivegradientlearningrule,avariantofthegeneralalgorithmofback-propagation[57], andaweightdecayfactorwhichinhibitedcomplexityofthemodels[58]. Thehistoricfirerecordsof fireandno-fireobservationsformodelbuilding(90%)werefurtherdividedintwo. Onepartwas usedforiterativetraining(70%,252cases)andtheotherpart(30%,108cases)forearlystopping,the periodicassessmentofperformanceaccuracyinordertoavoidlosinggeneralizationcapacitydueto overtraining[59]. Thecascade-correlationmodelsfollowedasimilarprocedureto[60,61],inwhichthe modelarchitecture(numberofnodesinthehiddenlayer)isoptimizedduringtraining. The best model found had an 8–6–1 (input–hidden–output) structure, and classification rates of 0.78–0.73–0.69 for training–test–validation datasets (Table 3). When selecting the best ANN classificationmodel,welookedforthehighestclassificationrateonobservedandpredictedfire/no-fire observations,balancedresultsbetweenthethreedatasetsandaparsimoniousarchitecture. Variables inthemodel,byorderofimportance,weredistancetoforesttracks(threetimesinputtothemodel), distancetourbanareas(twiceinputtothemodel),distancetopowerlines(twiceinputtothemodel), andpopulationdensity(onceinputtothemodel). Finally,thisbestfireoccurrencemodelwasrun at20-mresolutionpixelleveltogeneratetheignitionprobabilitygrid(IP;valuesrangingbetween0 and1;Figure4). Table3.Classificationtablewiththeresultsforthebestoccurrencemodel.Themodelwasgenerated withNeuralWorksPredict® v.3.30software. Thisoccurrencemodelwasusedtogeneratea20-m resolutionignitionprobabilitygrid(Figure4).Geospatialvariablesassociatedwiththehistoricalfire ignitions(200fireignitions,1985to2013firerecords;mapama.gob.es)(1)andarandomsamplewith thesamenumberofno-fireobservations(0)wereincludedwithinthefiremodelingdomain.Asetof 40cases(10%)wasusedforthevalidationofthemodel. ClassificationRate Class 0 1 Total 0.756 0 99 32 131 Training 0.802 1 24 97 121 0.779 Total 123 129 252 0.679 0 36 17 53 Test 0.782 1 12 43 55 0.731 Total 48 60 108 0.765 0 13 4 17 Validation 0.609 1 9 14 23 0.687 Total 22 18 40 Forests 2017, 8, 30   8 of 29  and a random sample with the same number of no‐fire observations (0) were included within the  fire modeling domain. A set of 40 cases (10%) was used for the validation of the model.  Classification Rate Class 0 1 Total 0.756  0  99  32  131  Training  0.802  1  24  97  121  0.779  Total  123  129  252  0.679  0  36  17  53  Test  0.782  1  12  43  55  0.731  Total  48  60  108  0.765  0  13  4  17  Validation  0.609  1  9  14  23  Forests2017,8,30 8of27 0.687  Total  22  18  40    Figure 4. Ignition probability grid generated with an artificial neural network using the geospatial  Figure4. Ignitionprobabilitygridgeneratedwithanartificialneuralnetworkusingthegeospatial variables  associated  with  the  observed  ignition  data  (1985  to  2013  historic  fire  records;  variablesassociatedwiththeobservedignitiondata(1985to2013historicfirerecords;mapama.gob.es). mapama.gob.es). This fire occurrence grid was used to generate the 10,000 fire ignition input data  Thisfireoccurrencegridwasusedtogeneratethe10,000fireignitioninputdatamaskedtoburnable masked to burnable fuels in the wildfire modeling domain. Unburnable areas (IP = 0) correspond  fuelsinthewildfiremodelingdomain.Unburnableareas(IP=0)correspondtourbandevelopment, to urban development, roads and water bodies.  roadsandwaterbodies. 2.2.3. Wildfire Spread and Behavior Simulation  2.2.3. WildfireSpreadandBehaviorSimulation We used FlamMap to simulate wildfires under conditions of constant fuel moisture, wind  WeusedFlamMaptosimulatewildfiresunderconditionsofconstantfuelmoisture,windspeed speed and wind direction [43]. We conducted five different weather scenarios at 20‐m resolution,  and wind direction [43]. We conducted five different weather scenarios at 20-m resolution, with with 10,000 wildfires per scenario (Table 2). FlamMap uses the two‐dimensional fire growth  10,000wildfiresperscenario(Table2).FlamMapusesthetwo-dimensionalfiregrowthminimumtravel minimum travel time algorithm (MTT) [62], which has been widely used worldwide at a broad  timealgorithm(MTT)[62],whichhasbeenwidelyusedworldwideatabroadrangeofscaleswith range of scales with multiple purposes [63–66]. The MTT algorithm replicates fire growth based  multiplepurposes[63–66]. TheMTTalgorithmreplicatesfiregrowthbasedontheHuygens’principle, on the Huygens’ principle, where the growth and behavior of the fire edge is modeled as a vector  where the growth and behavior of the fire edge is modeled as a vector or wavefront [62], and fire or wavefront [62], and fire spread distance is predicted by the Rothermel’s surface fire spread  spreaddistanceispredictedbytheRothermel’ssurfacefirespreadmodel[67]. Firedurationwassetat model [67]. Fire duration was set at 6 hour, in agreement with the active fire spread duration of  6hour,inagreementwiththeactivefirespreaddurationoftheobservedlargestwildfireeventsinthe the observed largest wildfire events in the study area (i.e., Juslapeña 2009). We did not consider  studyarea(i.e.,Juslapeña2009). Wedidnotconsiderbarrierstofirespreadorfiresuppressionefforts. barriers to fire spread or fire suppression efforts. Overall, modeled fires burned burnable pixels  Overall,modeledfiresburnedburnablepixelsatleastonceandmorethan100timesonaverage. at least once and more than 100 times on average.  FlamMapoutputsburnprobability(BP)andflamelengthprobability(FLP)grids,aswellasafire size(FS)textfileandthefireperimeters(polygons). Theburnprobability(BP)isthenumberoftimes apixelburnsasaproportionofthetotalnumberoffires,andisdefinedasfollows: BP=F/n (1) whereFisthenumberoftimesapixelburnsandnisthenumberofsimulatedfiresperrun(n=10,000 inthisstudy). Specifically,theconditionalburnprobabilityinthestudyareaistheBPgiventhatafire igniteswithinthefiremodelingdomainandspreadsfor6hoursatassumedfuelmoistureandweather conditions(97thpercentilefireweather). Fireintensity[68]isfirstpredictedbytheMTTalgorithm[62] andisconvertedintoflamelengthas: FL=0.0775×I0.46 (2) Forests2017,8,30 9of27 whereFLisflamelength(m)andI firelineintensity(kW·m−1 ,kW=kilowatt). Thentheprogram calculatesaFLPregularpointgrid(atthefiresimulationresolution)fromthemultipleburningfiresat differentflamelengths(i.e.,backing,headingandflankingfirespreadflamelengths). Foreverypixel intheFLPoutput,theprobabilityofflamelengthiscalculatedaticategoriesofdifferentfireintensity levels(FILs),giventhatatleastoneofthesimulatedfireshasburnedthepixel. Inthisstudy,FILswere obtainedastwenty0.5-mflamelengthcategories(forFIL -FIL andFIL >9.5m). 1 19 20 In the fire size (FS) text file output generated by FlamMap, the simulated burned area (ha) is attributed to each xy coordinate fire ignition. Moreover, we also obtained burned-area polygon shapefiles associated with each simulated fire and minimum travel time (MTT) major flow-paths polylineshapefilesforthefivefireweatherscenarios(Table2). Travelpathwaysarestraightlinesthat connectnodesandintersectcellstoformsegmentsforwhichfirebehavioriscalculatedfromtheinput data[43]. 2.3. ExpertJudgementofStructureSusceptibility Weusedaresponsefunction(RF)toapproximatestructuresusceptibility(potentiallosses)using fire intensity level model outputs [36]. To generate a customized RF for residential houses in the studyarea,weusedtheDelphimethod[69]. TheDelphimethodisaniterativequestionnaireprocess usedtoobtainareliableconsensusfromacarefullyselectedexpertpanel, andithasbeenusedin previous studies to determine wildfire causality from the personnel involved in fire suppression activities[70,71]. Weconductedaface-to-faceandanonymoustwo-roundquestionnaireprocesswiththeregional firefighting“BomberosdeNavarra”chiefs,focusingonthemostexperiencedinWUIfiresuppression incentralNavarra. Fireintensityisthemaincausativefactorofhomelossgiventhatafirereachesa housingstructure,andthereforeinthequestionnaire,potentialvaluelossofstructures(asapercentage) wasassociatedtofourdifferentfireintensityclassresponsefunctions(intensitylevelsofFIL ,FIL -FIL , 1 2 4 FIL -FIL ,andFIL -FIL ). Thefourfireintensityclasseswereselectedconsideringpreviousstudies 5 7 8 20 andthecapabilitiesofexistinggeospatialtoolstointegratethefiremodelingoutputswithpotential fireeffects[36,49]. Inthefirstroundofthequestionnaireprocess,theexpertsfilledthequestionnaire anonymouslyaccordingtotheirownpersonalexperiencetoreducetheeffectofdominantindividuals. Theninthesecondround,thequestionnairewasrepeatedtothesameexperts,butincludedresults from the first round (average values and deviation in the fire intensity classes) to meet a higher consensusandrefinethefinalresults. TheobtainedcustomRFpresentedmoderatetostronglosses in housing structures as fire intensity increased (Table 4), similar to RFs obtained in other studies conductedinMediterraneanareas[35]. Table4.Customresponsefunction(RF)usedtoapproximatefireeffectsintermsofvalueloss(%)on residentialhousesinthestudyarea[42]. Thefiremodelingoutputfireintensitylevels(FILs)were groupedintofourclassesforthegeospatialriskassessment[49].WeusedtheDeplhimethodtoobtain thesusceptibilityfunctionfromanexpertpanelcomposedofthemostexperiencedfirefighterchiefson wildlandurbaninterfacefiresuppression[69].Thewildfirehadnegativeimpactsinstructuresatall fireintensities. RelativeNetValueChange(%)atDifferentFireIntensityClasses Moderate Veryhigh Valuedasset Low(FIL1) (FIL –FIL ) High(FIL5–FIL7) (FIL –FIL ) 2 4 8 20 FL<0.5m 0.5m<FL<2m 2m<FL<3.5m FL>3.5m Residentialhouse −10 −45 −75 −95 Forests2017,8,30 10of27 2.4. ResidentialHouseEconomicValue WeusedtheofficialcadastermethoddescribedintheNavarraForalDecree334/2001ofNovember 26toassesstheeconomicvalue(V)oftheindividualhousingstructuresinthestudyarea. ThisForal Decree approves the procedure for the economic assessment of immovable property in the Foral CommunityofNavarrathroughouttheimplementationoftheComparisonMethodoftheaverage marketprices,withreferencetoInheritanceandGifttaxes,andoverPropertyTransferandCertified LegalDocuments(textpublishedintheBoletínOficialdeNavarraNo. 155of21December2001,and theBoletínOficialdeNavarraNo. 21of18February2002;lexnavarra.navarra.es). Themethodhas beenupdatedseveraltimessinceitsfirstpublication,withtheForalDecree39/2015of17Junebeing thelastupdate. Therearespecificmodelstoestimatethevaluesforflats,singleresidentialhouses, andparkingorstoragerooms. Weusedthemodelforsinglehouses,sincemostdwellingsinthestudy areawerewellpreservedruralhousesorrecentlybuiltconstructions. Themainparametersusedby themodelaretheyearoftheinformation,typeofindividualhouse,location,cadastralcategoryand conductedreforms,yearofconstruction,constructedsurface,andtheratioofconstructedsurfaceto urbandevelopmentpolygonsurface. Theresidentialhouseswithmorethanonecadastralsub-division (originalbuildinganddwellingexpansion)weremergedintoasingleunit. Weusedmarketprices from2015toobtainthemostup-to-datevalues(Table5). Table5.Summarytableofthecadastereconomicvalue(V)fortheresidentialhousesintheJuslapeña Valley.CouncilpolygoncadastercodesNo.3andNo.16donothaveresidentialhouses(Figure1A). Thecadastervaluewasestimatedfortheyear2015,consideringthemodelpublishedintheForal Decree334/2001ofNovember26(lexnavarra.navarra.es). Cadaster Council Residential CadasterEconomicValue(€) (PolygonNo.) Name Houses(No.) Average Median Maximum Minimum 1 Beorburu 11 108,825 109,767 151,432 52,124 2 Osacar 8 145,249 133,338 223,641 100,226 4 Osinaga 16 130,053 123,993 213,505 50,040 5 Aristregui 23 162,244 191,847 423,310 48,396 6 Larrayoz 17 156,733 138,493 269,403 77,480 7 Nuin 26 142,338 109,033 261,787 52,254 8 Marcalain 31 163,710 141,553 315,990 91,589 9 Iruzkun 1 132,192 132,192 132,192 132,192 10 Garciriain 12 139,194 143,079 199,405 74,682 11 Belzunce 48 168,233 135,984 483,043 64,657 12 Navaz 21 131,148 114,763 223,539 68,061 13 Ollacarizqueta 55 135,030 127,351 319,046 67,402 14 Unzu 16 154,063 174,413 218,129 88,067 15 Usi 21 111,784 108,407 167,685 71,229 2.5. Analysis Wildfiresimulationoutputswereusedtoassesslargefireinitiation,transmission,exposureand risktoresidentialhousesofruralcommunitieswithinthestudyarea(Figure2). Wecombinedfivesets offiresimulationoutputs(BP,FLP,FS,burnedareaperimetersandmajorflowpaths),oneforeach scenariobyweightingtherelativescenarioprobability(Table2).

Description:
Sassari 07100, Italy; [email protected]. 3 Keywords: wildland urban interface; wildfire simulation modeling; wildfire risk transmission; . We simulated wildfire spread and behavior (fire size, burned area polygons, flame length
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.